Article appearing in the February, 2003 edition of:
Systems Contractor News

Foxboro, MA: Should an emergency occur at Gillette Stadium, Foxboro, MA, the new home of the New England Patriots football team, authorities know that their evacuation messages over key sound systems will be projected with good intelligibility. In fact they insisted on it. In what could presage a trend for other public event spaces in the U.S., objective measurements were made on the sound systems to prove they could deliver a specified intelligibility rating.

Gillette Stadium

Ron Baker, principal of Dallas-based consulting firm, Wrightson, Johnson, Haddon and Williams, Inc. (WJHW), and designer of the stadium's sound systems, chose the Gold Line Model DSP30 Real Time Audio Analyzer with the STICistm Intelligibility Option to perform these tasks. The tests were the culmination of a sound system re-evaluation and, for two locations within facility, a major re-design process.

Baker related that the original sound system design specifications did not mandate the intelligibility requirements. WJHW had already completed that design, which had been awarded but not yet completed, when the authorities took a new look at evacuation requirements. They were particularly concerned about the main 68,000 spectator open-air seating bowl, and the two large club/lounges on opposite sides of the stadium.

Baker and his team fired up all the modeling tools at their disposal, like EASE, CATT, and Bose Modeler, to assess the original design, and to predict the performance, including intelligibility, of the proposed new one. "With the main seating area, we wound up making only minor changes, as we expected, but the clubs were a different story," Baker noted. "The clubs were originally designed with a fairly straightforward background/foreground music system, but as a result of the new direction, they were upgraded to a more elaborate system from both a sound level standpoint and a survivability standpoint."

Each 60,000 sq. ft. three-story club is 500 feet long and can hold up to 3000 people. These were challenging areas due to a full three-story glass wall, minimal absorption, an RT of 2 seconds, and high crowd noise. As a result, "we added 300 to 400 loudspeakers in those rooms," Baker said. "The primary proponent is to over-ride the crowd noise. The consensus [of code and other officials and design team members] was that once the announcements are underway, the crowd noise is expected to not exceed 80 to 85 dBA, so our system can produce 10 to 15 dB over that, with resulting levels over 100 dBA."

Baker added that the re-design also involved revised cable runs, system redundancies, and equipment placement. Components of the sound systems included EAW loudspeakers, BSS Audio Soundwebs for signal processing, and LynTec Crowd Compensating Computers for real-time ambient noise detection.

Each area has its own dedicated Soundweb, with one unit mounted at the main fire command center where the paging microphone is also located. The Soundwebs connet to a failure indicator panel with a toggle switch that allows bypassing the unit through a redundant path. Once the systems were installed and operational, WJHW personnel returned to the stadium to perform the intelligibility measurements, using the Gold Line Model DSP30, newly purchased for this project.

"We prepared a full test methodology, and submitted it for approval," Baker said. "We needed to meet 0.7 on a CIS (Common Intelligibility Scale). This corresponds approximately to 0.5 STI and about 12% ALcons."

The target rating, procedures and analyses were based on Appendix A of NFPA 72, 1999 edition which references IEC 60849, Testing Sound Systems for Emergency Purposes, that in turn refers to IEC 60268-16, Objective Rating of Speech Intelligibility by Speech Transmission Index.

"We tested in representative areas," Baker explained. "Each unique speaker and architectural configuration--sideline, end zone, overhangs, in the open--was examined with a total of 43 locations measured in the seating bowl. At every test location, the measurements were done six times and recorded on a spreadsheet."

Further analysis with standard deviation calculations were made to arrive at the resultant score. "Out of those 43 locations, we had one or two that barely failed, and with minor adjustments in EQ or level, we brought everything into compliance," Baker said.

The test setup for the clubs was the same, but with 34 locations tested. Overall, a total of 462 tests were made.

"To simulate actual conditions, the measurements were done with all parts of the sound system operating and with a recording of typical crowd noise at the level we were expecting," Baker related.

The test signal was supplied on the Gold Line STI-PAtm test tone CD Disk which produces the reference simulation of human speech. The test signal was developed at the TNO Laboratories, Human Factors Division (Perception/Speech Group) in the Netherlands.

The test signal is a combination of the audible spectra and modulation spectra of speech for different octave bands, based on a pending patent licensed to Gold Line by Bose Corporation, and utilizing a proprietary TNO methodology which involves compressing the number of bands, while maintaining accuracy, to produce faster measurements. Baker noted that each measurement took about 20 seconds.

"We are not aware of another system that does these measurements in this convenient of a fashion," Baker said. "I think the measuring equipment does provide a meaningful indication of intelligibility of the systems. It was also interesting to experience how slight adjustments (a couple of dB) in level or EQ could result in fairly significant changes in the result reported by the instrument."

And how did the measurements match up with the predictions? Baker said, "Our predictions for the clubs were close to what we actually measured and I expect that as we continue to do this work, we will be able to refine our predictions with more real-world experience.

Baker also expects other venues will start to insist on intelligibility guarantees.

"The IEC standard is not an official part of the fire code as now it is an appendix, but it could appear in the main body of the code the next time it's revised," Baker cautioned. "These types of measurements are very prevalent in Europe, and already required in England, and it's only a matter of time that we'll see them required here [in the U.S.]"

Home DSP30 & STI-PAtm